Use OpenAI's most capable GPT Image 2.5 model for precise generation and demanding image edits. Build polished product, campaign, and design assets with detailed prompts, multiple references, and explicit output controls.
默认已选择 GPT Image 2.5 Sunburst。你可以在这里通过提示词和可选参考图生成单张图片；需要 Variations、CSV 或 Products 时，可进入批量工作台。
Model overview
Sunburst is OpenAI's most capable image generation and editing model, intended for precise, premium workflows where fidelity matters more than the shortest wait.
OpenAI describes Sunburst as the stronger choice for creators, agencies, brands, and developers who need refined product imagery, controlled campaign assets, or sophisticated edits. It is designed to follow complex visual instructions and preserve important reference details more reliably.
In PixBulk, Sunburst uses the same practical workspace as Flare: text prompts, multiple reference images, quality levels from Low through Max, three canvas shapes, background behavior, output format, and multi-image generation. Its value appears when the brief includes tight composition, identity, product, typography, or edit constraints.
Sunburst is built for briefs where several details must work together and a small error can undermine the final asset.
Create invitations, posters, product cards, packaging concepts, and information-rich layouts with explicit copy, hierarchy, spacing, ornament, and material direction. Always proofread generated text before delivery.
Use references to anchor identity, products, objects, or visual language while changing another part of the scene. Tell the model which attributes are immutable and which can be reinterpreted.
Combine setting, pose, wardrobe, camera, light, material, color, and graphic constraints in one brief. Sunburst is the better family member when those instructions must be reconciled with high visual fidelity.
Escalate an approved direction to Sunburst when the final result needs careful materials, believable lighting, controlled brand details, and fewer compromises across the composition.
Sunburst is designed for editing precision. Supply the source and any supporting references, identify the protected details, then describe the exact replacement, addition, removal, restyling, or environmental change.
Keep the person, facial identity, pose, expression, and crop unchanged. Replace only the environment with an elegant reception setting, match the original depth of field, and add warm practical lights with physically believable reflections.
Choose Sunburst when the deliverable is close to final and the cost of an imprecise edit or missed instruction is higher than the value of faster iteration.
Develop controlled hero shots, campaign settings, packaging concepts, and material-rich product scenes from a detailed brief and reference set.
Translate an approved concept into refined assets where the composition, palette, visual identity, copy, and subject treatment must feel intentional together.
Change one part of an existing visual while protecting identities, products, poses, or other important elements named in the instruction.
Treat the prompt as an art-direction document: define the goal, establish constraints, and separate protected details from requested changes.
Goal + subject + protected details + requested changes + composition + camera + lighting + materials + exact text + output context
Create a vertical luxury skincare campaign featuring the supplied amber bottle, preserve its exact silhouette, cap, label colors, and logo placement, center it on wet charcoal stone, close three-quarter camera angle, cool dawn backlight with warm rim reflections, subtle mist and water beads, generous negative space above for campaign copy, photoreal materials, restrained premium art direction.
For edits, list protected identity, product, pose, framing, and brand details before explaining the transformation.
Say which image provides the subject, product geometry, palette, composition, texture, or environmental direction.
Use position, scale, depth, crop, negative space, lens, and viewing-angle language to make the intended composition testable.
Explore at a practical tier, then increase quality after the composition and art direction are stable enough to justify the added credits.
同一套三步流程既适合单次创作，也适合结构化批量生产。
页面加载时已默认选择 GPT Image 2.5 Sunburst。输入一个提示词，按需添加支持的参考图；更大的任务可进入批量工作台。
描述主体、场景、构图、风格与限制，然后选择 GPT Image 2.5 Sunburst 支持的比例和分辨率；只有模型支持时才添加参考图片。
提交任务并比较 GPT Image 2.5 Sunburst 结果，再通过有针对性的提示词修改重新生成。完成的作品会保存在历史记录中，便于后续查看和下载。
These are the settings currently exposed by the PixBulk integration. The generator remains the source of truth if provider capabilities change.
| Setting | Available choices |
|---|---|
| Quality | Low, Medium, High, XHigh, Max |
| Canvas shape | Square (1024 × 1024), portrait (1024 × 1536), landscape (1536 × 1024) |
| Background | Auto, opaque, transparent |
| Output format | PNG, JPEG, WebP |
| Images per request | 1–4 images |
| Reference images | Multiple images, subject to the current model input limit |
Sunburst and Flare expose the same core controls in PixBulk. Sunburst is the precision-first option; Flare is the faster default for ideation, everyday generation, and larger iteration loops.
| Model | Best for | Primary tradeoff |
|---|---|---|
| GPT Image 2.5 Flare | Rapid concepts, social assets, product drafts, high-volume creative iteration | Optimized for speed rather than the family's highest precision |
| GPT Image 2.5 Sunburst | Precise edits, premium product imagery, campaigns, demanding final assets | More generation time for greater capability and control |
Use the model as part of a reviewed creative workflow. Quality settings improve rendering effort, but they do not replace a clear brief or human approval.
Sunburst can take longer than Flare. Use it when precision is worth the wait, not automatically for every early concept.
Even strong reference preservation is not a guarantee. Verify faces, products, logos, labels, geometry, and any detail that must match a source exactly.
Complex prompts can contain conflicting instructions. Prioritize constraints, remove ambiguity, and split a difficult transformation into focused passes when needed.
PixBulk exposes Low, Medium, High, XHigh, and Max. OpenAI also documents an Auto option, but it is intentionally not part of the current PixBulk selector.
关于访问方式、积分、输入、输出控制、批量工作流与审核的实用回答。
能力说明来自官方供应商资料与 PixBulk 集成所使用的 API 文档。应用界面可能有意只开放其中经过支持验证的子集。
打开上方默认选择 GPT Image 2.5 Sunburst 的生成器，编写简报并选择交付格式，在同一工作区中把想法变成可审核结果。
使用此模型生成
